Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Nov 10.
Published in final edited form as: Cortex. 2025 Jul 24;191:1–11. doi: 10.1016/j.cortex.2025.07.004

Beyond the Cortex: Cerebellar Contributions to Apraxia of Speech in Stroke Survivors

Makayla Gibson 1, Alberto Maydeu-Olivares 1, Roger Newman-Norlund 1, Christopher Rorden 1
PMCID: PMC13746207  NIHMSID: NIHMS2208437  PMID: 40752269

Abstract

Apraxia of Speech (AOS) is a motor speech disorder affecting speech planning and coordination, often co-occurring with aphasia after left hemisphere stroke. While commonly linked to cortical damage, the cerebellum’s role remains underexplored. This study examined associations between cerebellar gray matter volume and AOS severity in individuals with chronic left hemisphere cortical stroke.

Participants from the Aphasia Recovery Cohort (N = 142) completed structural MRI and AOS assessment using the Apraxia of Speech Rating Scale. Regional cerebellar gray matter volumes were extracted from T1-weighted scans using MATLAB, SPM12, CAT12, and the SUIT atlas, adjusted for total intracranial volume. Multiple linear regressions and partial correlations tested relationships between cerebellar volume and AOS severity, controlling for lesion volume, age, and time post-stroke.

Reduced volume in right cerebellar regions was significantly associated with greater AOS severity. Right lobule V was the strongest predictor (R2 = .167, β = −0.346, p < .001), followed by lobule VI (R2 = .130, β = −0.290, p < .001) and vermis VIIb (R2 = .117, β = −0.262, p < .001). Right hemisphere volume overall was more predictive (R2 = .116) than the left (R2 = .080). Dysarthria models showed weaker associations, indicating specificity to AOS.

These findings demonstrate that cerebellar structural integrity, particularly in the contralateral right hemisphere, contributes to AOS severity models. Results support expanding AOS models beyond cortical regions and suggest cerebellar-targeted strategies may enhance post-stroke speech rehabilitation models.

Keywords: apraxia of speech, cerebellum, stroke, aphasia, magnetic resonance imaging, speech disorders

Introduction

Apraxia of Speech (AOS) disrupts an individual’s ability to plan and coordinate speech movements, creating a profound disconnect between language ability and speech production capacity. Affecting 30–58% of left hemisphere stroke patients, AOS commonly occurs alongside other aphasic symptoms, presenting a significant clinical challenge in stroke rehabilitation.1 While distinct from the language deficits seen in aphasia or the muscle weakness characteristic of dysarthria, AOS specifically impairs the programming of speech movements, resulting in inconsistent articulation and prosody.2 While lesion mapping studies of AOS have identified the role of injury to left hemisphere regions,3 it is well-established that these features explain only a moderate proportion of the variability in symptom severity. For example, Levy and colleagues (2024) 4 found that lesion load accounted for 38–59% of variance among patients with aphasia, across timepoints and domains. This leaves room for other brain regions, including those indirectly impacted by cortical injury, such as the cerebellum, which may help explain variability in outcome and offer further insight into AOS.

Understanding the neural basis of AOS requires examining brain structures beyond traditional language regions, with particular attention to the cerebellum’s role in speech motor control.5 Once regarded primarily as a center for motor coordination, the cerebellum is now recognized as a critical component of speech processing.6 Through extensive connections with cortical and subcortical regions, the cerebellum plays a vital role in motor planning and execution, regulating movement precision while ensuring the smooth execution of complex speech-related motor actions, with neuroimaging studies consistently demonstrating cerebellar activation during speech tasks, indicating involvement extends well beyond basic motor coordination.7,8

The cortico-cerebellar loop facilitates the cerebellum’s role in speech motor control through specific pathways connecting cortical speech regions with cerebellar structures. Corticopontine fibers project from Broca’s area, supplementary motor area, primary motor cortex, and prefrontal areas to the pontine nuclei. Information is then relayed via pontocerebellar fibers to the cerebellar cortex, where the cerebellum processes this input and projects back to the cortex.8 These pathways are crucial for refining motor commands, predicting sensory outcomes of speech movements, and enabling real-time adjustments of articulatory patterns, and particularly during novel or complex speech tasks.9,10 Both functional neuroimaging and lesion studies have revealed the cerebellum’s involvement in feedforward planning and feedback control mechanisms, both essential for articulation.11 Given the cerebellum’s strong connectivity with language networks, it should come as no surprise that targeting this structure via transcranial direct current stimulation (tDCS) has also been shown to improve language outcomes among patients with chronic aphasia.12

The clinical relevance of cerebellar function in speech disorders is most evident in ataxic dysarthria, resulting from lesions within the cerebellar hemispheres or superior cerebellar peduncles. More specifically, in ataxic dysarthria we see consistent disruptions in the coordination, timing, and precision of speech movements, leading to hallmark features such as irregular articulatory breakdowns, distorted prosody, and excessive variability in speech rate and rhythm. These deficits arise from the impaired cerebellar modulation of motor output, which is essential for the smooth execution of speech.13,14, However, unlike ataxic dysarthria, which is a hallmark of direct cerebellar lesions,14 AOS is typically associated with cortical lesions, particularly in the dominant left hemisphere. Lesion-symptom mapping studies have consistently shown that AOS typically arises from damage to left hemisphere areas such as the inferior frontal gyrus, premotor cortex, and supplementary motor area.3,15 These regions, crucial for generating and sequencing speech motor commands, are also commonly affected in individuals with non-fluent aphasia,16 and their disruption leads to the hallmark symptoms of AOS: articulatory groping, speech sound distortions, and abnormal prosody, but differ from ataxic dysarthria by being primarily a disorder of motor planning and programming rather than pure execution. While ataxic dysarthria results from impaired coordination of already formulated motor commands, AOS is characterized by difficulty in selecting and sequencing the correct speech movements, often leading to trial-and-error articulatory groping, inconsistent errors, and increased difficulty with longer or more complex utterances.17

While significant research effort has been dedicated to mapping cortical contributions to language recovery after stroke,18–23 fewer studies have examined the cerebellum’s role in these outcomes. However, a recent investigation identified associations between cerebellar atrophy and speech impairment in a chronic post-stroke aphasia group, suggesting that cerebellar degradation, potentially arising from cross cerebellar diaschisis or disrupted connectivity, may exacerbate language deficits when cortical structures are compromised.24 Advanced imaging techniques, such as voxel-based morphometry (VBM) and diffusion tensor imaging (DTI), further support this link by revealing structural changes in the cerebellum following cortical stroke, although fewer studies have explicitly investigated whether cerebellar changes contribute to motor speech impairments such as AOS.25,26 Notably, the cerebellum has been excluded from neuroimaging analyses, which has hindered progress in understanding the entire picture of the cerebellum’s role in cognition and clinical disorders. 27

These secondary cerebellar effects, resulting from cortical deafferentation rather than primary cerebellar lesions, may independently contribute to chronic deficits in speech motor control. In this study, we focus on the cerebellum’s role in chronic AOS among individuals with left hemisphere cortical stroke, investigating the independent contribution of cerebellar structural integrity to AOS severity.

Despite growing evidence for cerebellar involvement in speech motor control 5,28 and mounting research on post-stroke language outcomes, a critical knowledge gap remains: How does cerebellar structural integrity impact motor speech impairments, and more specifically AOS severity, in left hemisphere stroke. Our study addresses this gap by examining how cerebellar grey matter volume relates to AOS severity in chronic aphasia patients with left hemisphere cortical lesions. By focusing on the chronic stage of recovery, we can examine long-term structural adaptations, providing insight into how non-lesioned brain regions contribute to speech motor control.

By accounting for both cortical lesion maps and cerebellar grey matter volumes, our analysis aims to isolate the contribution of cerebellar atrophy, likely driven by cortical disconnection (cross-cerebellar diaschisis), to persistent AOS symptoms, in the absence of direct cerebellar damage. By associating the integrity of the cerebellum (as assessed with structural MRI scans) with the Apraxia of Speech Rating Scale,29,30 our investigation seeks to advance both theoretical models of motor speech disorders and practical approaches that might leverage the cerebellum’s adaptive capabilities. This research may identify potential markers for speech recovery and inform targeted rehabilitation strategies, representing a significant shift from traditional cortical-focused studies of AOS.

Materials & Methods

Participants

Data included in the following investigation were from the Aphasia Recovery Cohort, where participants were recruited from various studies, and provided informed consent approved by the Institutional Review Board, located in Columbia, SC, at the University of South Carolina (USC). To be eligible for enrollment, individuals had to have experienced a left-hemisphere stroke at least six months prior, be between 21 and 80 years old, and have no contraindications to MRI or additional neurological conditions (e.g., multiple sclerosis, Parkinson’s disease, dementia).31 In this cohort, 142 unique individuals underwent scanning (57 females, 85 males; 29–80 years old, M = 61, SD = 10), and 12 participants reported left-handedness prior to their stroke. On average, imaging was conducted 5.1 years post-stroke (SD = 4.7 years). Some participants completed multiple speech assessments over time, resulting in a total of 178 assessment data points (M = 4.5 years post-stroke, SD = 4.3 years). For participants who completed multiple speech assessments (M = 1.25 assessments per participant), the same imaging data was used across sessions, with the analysis accounting for repeated measures. Based on ASRS scores (range: 0–21, M = 4.6, SD = 5.9), 39 participants scored ≥8, consistent with a diagnosis of apraxia of speech. No participants were excluded based on aphasia severity, and the inclusion of a full spectrum of impairment ensures the validity of our findings.

Assessments

To evaluate AOS severity, the Apraxia of Speech Rating Scale 30 was administered, due to its well-established validity, reliability, and sensitivity in diagnosing AOS.29,30 This assessment measures various speech production parameters, including articulation, prosody, and fluency, generating a severity score for AOS, dysarthria and aphasia. In addition to the AOS score, we utilized the dysarthria score from the same assessment to investigate the specificity of AOS in the cerebellum. Given that some participants completed multiple assessments at different time points, statistical models accounted for repeated measures to ensure robust analyses.

The ASRS was performed at the same time and by the same clinician, allowing for a direct comparison between AOS and dysarthria severity. If cerebellar atrophy was simply a consequence of left hemisphere lesions rather than being specifically associated with AOS, we would expect cerebellar regions predictive of AOS severity to similarly predict dysarthria severity.

Neuroimaging

Structural MRI scans were acquired for all participants at the McCausland Center for Brain Imaging. High-resolution T1 images were acquired using a Siemens Trio/Prisma 3T scanner equipped with a 20-element head-neck coil with the following parameters: an MP-RAGE sequence with 1 mm isotropic voxels, a 256 × 256 matrix size, a 9-degree flip angle, and a 192-slice sequence with a short repetition time = 2250 ms, inversion time = 925 ms, and echo time = 4.11 ms.

Neuroimaging Analysis

Cerebellar volume and total intracranial volume (TIV) were calculated using high-resolution T1-weighted structural MRI images. All images were processed using MATLAB (Mathworks, 2022b), SPM12 version 778132 and the CAT12 (v12.6, 1700) VBM toolbox with default settings.33 T1-weighted structural images underwent enantiomorphic healing prior to processing to ensure accurate spatial normalization.34,35 Cerebellar volume metrics were extracted using the default CAT12 processing pipeline and adjusted for each participant’s total intracranial volume (TIV) to account for individual differences in brain size.36 This normalization step converted the volume measurements from cubic centimeters to percent total brain volume, providing a unitless metric that compensates for variability in TIV across participants. 36 Gray matter volumes, expressed as percentages, were calculated separately for the left and right cerebellar hemispheres, as well as for each of the 28 spatially distinct main lobules regions defined by the SUIT cerebellar atlas.37

Chronic stroke lesions were manually delineated by trained raters using MRIcroGL12 software.38 Binary NIFTI lesion masks, aligned to native space, were created with lesioned voxels labeled as ‘1’ and non-lesioned voxels as ‘0.’ Lesion volume was quantified by summing the number of voxels labeled as ‘1’ and converting this value to cubic millimeters based on voxel dimensions in the native image space. This approach ensured accurate volume measurements while preserving spatial alignment.

To visualize the spatial distribution of lesions across the cohort, an overlay map was generated by summing all individual lesion masks and normalizing the values to reflect lesion overlap. The resulting lesion map (Figure 1) highlights regions with the highest lesion frequency, where darker blue areas (value = 10) indicate minimal overlap and brighter red areas (value = 89) represent the most commonly affected regions. This visualization provides a comprehensive overview of lesion distribution, facilitating the interpretation of lesion burden across participants.

Figure 1:

Figure 1:

Group lesion distribution map displayed in neurological convention. The slices correspond to the MNI coordinate space at z-dimension intervals of −35, −24, −13, −3, 8, 18, 29, 39, 50, and 60. Lesion intensity is represented by a color gradient, with darker blue areas (value = 10) indicating minimal overlap and brighter red areas (value = 89) representing the most commonly affected regions, with the gradient showing overlap from 10 to 89 individuals.

Statistical Analysis

A multiple linear regression was used to investigate whether cerebellar grey matter volume improves the prediction model of AOS severity, beyond what is predicted by lesion volume, age and days post-stroke at the time of AOS assessment. Given that some participants had multiple speech assessments, days post-stroke at the time of assessment was included as a covariate to account for repeated measures while maintaining statistical independence of neuroimaging data.

We conducted both linear regression and partial correlation analyses, because while both partial correlation and linear regression account for the same control variables, they provide distinct but complementary insights. Partial correlation measures the direct association between two variables while holding other variables constant, offering a clearer view of their relationship, while regression extends this by modeling the unique contribution of each predictor, accounting for multiple covariates and their combined effects on the outcome.

AOS scores, lesion volume and days post-stroke were positively skewed, so to improve model fit in the data, log transformations were applied to variables with right-skewed distributions. AOS severity scores were log-transformed using a log(x + 1) approach to accommodate potential zero values.39 Age at scan was retained in its original scale, as its distribution was approximately normal.

To assess the specificity of cerebellar contributions to AOS severity, we conducted parallel analyses using the dysarthria score as the dependent variable, rather than AOS score. Dysarthria severity scores from the ASRS assessment were also log-transformed using a log(x + 1) approach to improve normality. Pearson correlation analyses were performed to examine the relationships between cerebellar gray matter volume and dysarthria severity. Additionally, a multiple linear regression model was used to determine whether cerebellar gray matter volume predicted dysarthria severity while controlling for lesion volume, days post-stroke, and age at scan. This analysis allowed us to evaluate whether observed associations with cerebellar GMV were specific to speech motor planning deficits or reflective of broader articulatory impairment, or even an overall cross-cerebellar diaschisis following stroke.40

Results

Linear Regression

A multiple linear regression model was used to examine the relationship between cerebellar gray matter volume and AOS severity, while controlling for lesion volume, days post-stroke, and age at scan. The baseline model, which included only these covariates, accounted for 5.4% of the variance in AOS severity (R2 = .054). To assess the contribution of individual cerebellar regions, each region of interest (ROI) was added independently to the model, and the change in R2 was evaluated.

Several cerebellar regions significantly improved model fit (figures 2 & 3). The strongest predictor was right lobule V, which explained an additional 11.3% of the variance (R2 = .167, β = −0.346, p < .001). Right lobule VI also demonstrated a robust association with AOS severity, contributing an additional 7.6% of the variance (R2 = .130, β = −0.290, p < .001). The vermis, particularly VIIb, was also a significant predictor, increasing the explained variance by 6.3% (R2 = .117, β = −0.262, p < .001). Additional ROIs, including right Crus I (R2 = .104, β = −0.238, p = .003) and the vermis of Crus II (R2 = .102, β = −0.223, p = .004), also significantly contributed to explaining AOS severity. A summary of the effects of covariates, including age, lesion volume, and days post-stroke, across these models is provided in Table 1.

Figure 2 :

Figure 2 :

Flatmap of cerebellar ROIs derived from the SUIT atlas, depicting the contribution of GMV to predicting AOS severity. ROIs shown in light yellow indicate the highest R2 values, reflecting greater GMV contributions, while blue regions represent the lowest R2 values. Only statistically significant ROIs (p < 0.05) are colored, with non-significant regions shown in grey. Each ROI was included as an independent predictor in the model, which controlled for age at scan, days post-onset, and lesion volume.

Figure 3 :

Figure 3 :

Bar graph comparing R2 values from linear regression models predicting AOS and dysarthria severity using cerebellar GMV within each ROI. Only ROIs with at least one statistically significant model (p < 0.05) are shown. Bars are color-coded: green for AOS, purple for dysarthria, and light gray for non-significant models.

Table 1:

Summary of Covariate Effects on AOS Severity Across Regression Models of Significant Cerebellar Regions

Model DV Covariate β (Standardized) p-value R2 (Model)
M0 AOS Score Age at Scan −0.060 .426 .054
Days Post-Stroke −0.011 .889
Lesion Volume 0.225 .004
M1 AOS Score Right Lobule V −0.346 <.001 .167
Age at Scan −0.135 .066
Days Post-Stroke −0.015 .833
Lesion Volume 0.246 <.001
M1 AOS Score Right Lobule VI −0.290 <.001 .130
Age at Scan −0.151 .051
Days Post-Stroke −0.007 .924
Lesion Volume 0.225 .003
M1 AOS Score Vermis VIIb −0.262 <.001 .117
Age at Scan −0.132 .085
Days Post-Stroke −0.028 .711
Lesion Volume 0.208 .006
M1 AOS Score Right Crus I −0.238 .003 .104
Age at Scan −0.131 .085
Days Post-Stroke −0.024 .711
Lesion Volume 0.257 <.001
M1 AOS Score Vermis Crus II −0.223 .004 .102
Age at Scan −0.098 .195
Days Post-Stroke −0.015 .844
Lesion Volume 0.247 .001

When considering the summed cerebellar ROIs of each hemisphere, right cerebellar volume proved stronger predictive value than the left. The summed right cerebellar regions significantly improved the model (R2 = .116, β = −0.258, p < .001), whereas the summed left cerebellar regions had a smaller but still significant effect (R2 = .080, β = −0.172, p = .031). These findings suggest that reduced gray matter volume in specific right cerebellar regions, particularly lobules V and VI, is strongly associated with greater AOS severity, though partial volume effects may contribute signal from adjacent white matter.

Pearson Correlation

To further explore these relationships, a correlation analysis was conducted to assess the association between cerebellar integrity and AOS severity, while controlling for lesion volume, days post-stroke, and age at scan. The results demonstrated significant negative correlations between AOS severity and several cerebellar regions, reinforcing findings from the regression analysis. The strongest correlations were observed in the right cerebellum, particularly right lobule V (r = −0.347, p < .001), right lobule VI (r = −0.283, p < .001), and vermis VIIb (r = −0.258, p < .001), indicating that greater atrophy in these regions was associated with more severe AOS symptoms. Additionally, the sum of all right cerebellar ROIs was significantly correlated with AOS severity (r = −0.257, p < .001), further emphasizing the right cerebellum’s role in speech motor control.

Other notable correlations included right crus I (r = −0.231, p = .003), vermis crus II (r = −0.225, p = .004), right VIIb (r = −0.223, p = .004), and right VIIIa (r = −0.223, p = .004). In contrast, while some left cerebellar lobules also showed significant correlations with AOS severity, the effects were weaker. Left lobule V (r = −0.196, p = .012), left VI (r = −0.170, p = .028), and left crus I (r = −0.189, p = .015) all reached statistical significance, though with smaller effect sizes compared to their right-hemisphere homologues. The sum of all left cerebellar ROIs also showed a significant, albeit weaker, correlation with AOS severity (r = −0.167, p = .031). These relationships can be seen in figure 4.

Figure 4 :

Figure 4 :

Heatmap displaying Pearson’s correlation coefficients between cerebellar ROI GMV and clinical measures of AOS and dysarthria. Only statistically significant correlations (p ≤ 0.05) are shown; non-significant values are rendered transparent. Correlation strength is color-coded, with the strongest negative associations shown in yellow (r = –0.35) and the weakest in purple (r = –0.15), emphasizing the gradient of negative relationships.

Several cerebellar regions, including vermis crus I, left crus II, left VIIb, left VIIIa, left lobule VIIIb, left IX, right IX, left X, and vermis X, did not reach statistical significance (p > .05), suggesting minimal involvement in AOS severity.

Additional Analysis: Specificity of Cerebellar Contribution to AOS Severity

To ensure that the observed relationship between cerebellar gray matter volume and AOS severity was specific to apraxia, and not merely a reflection of global cerebellar atrophy in response to left hemisphere damage, we conducted a control analysis using dysarthria scores from the same ASRS assessment as an alternative dependent variable.

However, the linear regression model predicting dysarthria severity showed limited explanatory power, with an initial R2 of only .014 when controlling for lesion volume, time post-stroke, and age. When cerebellar ROIs were added independently of one another, only a few regions showed weak associations with dysarthria severity. Notably, the right lobule V (β = −0.208, p = .008, R2 = .055), right lobule VI (β = −0.224, p = .005, R2 = .060), right Crus I (β = −0.169, p = .038, R2 = .040), and right VIIb (β = −0.155, p = .050, R2 = .037) were significantly associated with dysarthria severity, however, the overall explained variance remained low. (Figure 3).

Pearson correlations (Figure 4) further confirmed these weak relationships, with the strongest association being right lobule VI (r = −0.215, p = .005). In contrast, the sum of all left and right cerebellar ROIs (suml, sumr) showed only marginal non-significant trends (r = −0.052, p = .507 for suml; r = −0.149, p = .055 for sumr).

Multiple Comparisons

While each cerebellar ROI was entered into its own regression model to evaluate regional specificity, we acknowledge that conducting multiple statistical tests increases the risk of false positives. To address this, we applied both False Discovery Rate (FDR) and Bonferroni corrections across all ROI-level models. Following FDR correction, several regions remained statistically significant, including right lobule V (adjusted p = 0.006), right lobule VI (adjusted p = 0.006), vermis VIIb (adjusted p = 0.006), right crus I (adjusted p = 0.013), and right sum (adjusted p = 0.006). These regions continued to demonstrate strong associations with AOS severity beyond chance. Notably, right lobule V, right lobule VI, vermis VIIb and right sum also remained significant under the more conservative Bonferroni correction (adjusted p = 0.032), providing further evidence of their robust contribution.

Controlling for Vascular Lesion Distribution

To assess whether the association between cerebellar GMV and AOS could be confounded by lesion location in the cortex, we conducted additional linear regression analyses that included the percent damage to 16 vascular territories in the left hemisphere based on the Arterial Territories Atlas.41 These models also included age at scan, lesion volume, and days post-stroke. Even after controlling for lesion distribution, several cerebellar regions continued to significantly explain unique variance in AOS severity. For example, right lobule V remained a strong predictor of AOS (p < .001), improving model fit from R2 = .252 (lesion distribution, lesion volume, age, days post-stroke model) to R2 = .329. Significant vascular covariates in this model included damage to the frontal pars of the middle cerebral artery (p = .001) and the temporal pars of the posterior cerebral artery (p = .027). Similar patterns were observed for right lobule VI (p < .001, R2 = .321), vermis VI (p = .003, R2 = .308), and vermis VIIIa (p < .001, R2 = .322), each contributing explained variance beyond the vascular model. These results suggest that cerebellar atrophy offers explanatory power for AOS severity that cannot be accounted for solely by cortical lesion burden.

Accounting for Co-Occurring Aphasia

Given the known frequent co-occurrence of apraxia of speech and aphasia following left hemisphere stroke,1 it is important to consider whether general language impairment may influence AOS severity ratings. To address this, we conducted follow-up regression analyses that included aphasia severity (from the ASRS total aphasia score) as a covariate alongside age, lesion volume, and days post-stroke. As expected, aphasia severity significantly improved the model and accounted for a substantial portion of the variance in AOS severity (R2 = .45), reflecting shared clinical features across these disorders.

However, even after controlling for aphasia severity, several cerebellar regions remained significant independent predictors of AOS severity. Most notable in these analyses, Right Lobule V (p < .001, R2 = .497) and Right I–IV (p = .004, R2 = .478) continued to show robust associations. These findings suggest that cerebellar gray matter volume contributes unique explanatory value to motor speech planning deficits, beyond what can be attributed to aphasia severity, lesion volume, or stroke chronicity. While aphasia contributes significantly to overall speech disruption, our results clearly demonstrate that cerebellar atrophy explains additional and distinct variance in AOS severity further reinforcing its role in speech motor planning after stroke.

Discussion

Cerebellar Volume & AOS Severity

Our findings reveal a robust relationship between cerebellar gray matter volume and AOS severity in individuals with chronic left hemisphere stroke, with the strongest associations localized to right cerebellar regions, even after controlling for total intracranial volume, lesion volume, chronicity, and age. While our analyses focused on gray matter using established segmentation tools, the spatial resolution of structural MRI does introduce the possibility that some effects may also reflect changes in nearby white matter. Even so, these results underscore the cerebellum’s critical contribution to speech motor planning and execution, extending our understanding of AOS beyond traditional cortical models.

Relationship to Ataxic Dysarthria

A valuable consideration here is the complex relationship between cerebellar dysfunction, speech motor disorders, and cortical stroke. Our findings complement the existing literature on ataxic dysarthria, that highlights the cerebellum’s critical role in speech motor control.5,28 Unlike AOS, which is characterized by impaired planning and programming of speech movements, resulting in inconsistent articulation errors and disrupted prosody, ataxic dysarthria emerges directly from cerebellar lesions. More specifically, ataxic dysarthria resulting from the cerebellar insult presents as consistent irregular articulatory breakdowns, distorted prosody, and variability in speech rate and rhythm.14 While our study focuses on AOS in the context of left hemisphere cortical stroke, it is important to recognize that the cerebellum’s role extends beyond motor coordination to include aspects of speech motor planning. Consequently, cerebellar atrophy, particularly that occurs as a secondary consequence to cortical lesions, can disrupt both the execution and planning of speech movements, potentially contributing to chronic AOS in these individuals. This functional overlap underscores the cerebellum’s integral contribution to the broader networks that support speech production.

Heterogeneity in Stroke Outcomes

Critically, not all individuals with left hemisphere stroke in our cohort were diagnosed with AOS, reinforcing the variability in speech motor outcomes following cortical injury. The heterogeneity of stroke outcomes reflect the complex networks underlying speech production. Factors such as lesion location,20,42–44 individual neural plasticity, 45,46 pre-existing compensatory mechanisms 47,48 and variability of age related cerebellar degeneration 49 contribute to variability in speech motor impairments. This heterogeneity underscores the importance of individualized approaches to understanding and treating post-stroke speech disorders.

Lesion Volume & Cerebellar Gray Matter Volume

Another intriguing aspect of our analysis that is worthy of discussion here, is the lack of correlation between lesion volume and cerebellar grey matter volume. Specifically, right cerebellar gray matter volume (sumr) was not significantly correlated with cortical lesion volume when controlling for age and days post-stroke (r = –0.037, p = 0.640), suggesting that cerebellar structural changes reflect more than overall lesion burden, and cerebellar structural changes are not merely a passive response to cortical damage, but potentially represent neural reorganization or specific pathophysiological processes related to speech motor control.

Structural vs. Functional Imaging Approach

Our methodological approach of using structural MRI scans warrants discussion. While functional neuroimaging might provide insights into neural activation patterns, structural imaging offers a perspective of long-term tissue adaptations.50,51 Structural scans reveal grey matter volume changes that reflect neuroplastic reorganization and even potential compensatory mechanisms in these chronic stroke patients. This approach allows us to examine how underlying neural architecture relates to speech motor deficits, providing insights that dynamic functional imaging might not capture.

Limitations and Future Directions

Despite the robust associations identified, this model could be improved with complementary techniques. Our analysis relied on neuroimaging data collected at a single time point, limiting insights into cerebellar structural changes over time. While we controlled for lesion volume, we did not classify participants based on specific lesion locations: voxel-based lesion-symptom mapping would clarify whether damage to specific cortical regions influences cerebellar structure and AOS. Additionally, our study lacked diffusion tensor imaging measures to examine white matter tract integrity between the cerebellum and cortical regions, and while the ASRS provides comprehensive AOS assessment, incorporating additional measures would offer a more complete picture of cerebellar structure’s relationship to various aspects of speech and language function.52,53 Finally, while the SUIT atlas is limited by the resolution constraints of T1-weighted MRI in distinguishing cerebellar gray and white matter, it provides a widely used cerebellar segmentation framework (cited >900 times),54 offering a reliable and reproducible method for ROI-based analysis.

Conclusion / Summary

This study provides compelling evidence that the cerebellum’s integrity predicts chronic speech motor planning deficits following left hemisphere stroke. Our findings demonstrate that reduced gray matter volume in specific cerebellar regions, particularly in right lobules, is significantly associated with AOS severity in chronic stroke patients. These relationships persist even after controlling for total intracranial volume, lesion volume, time post-stroke, and age, suggesting that cerebellar involvement represents a specific mechanistic contribution to speech motor planning deficits rather than merely reflecting crossed cerebellar diaschisis following left hemisphere damage. The observed structural cerebellar differences likely represent critical adaptations that directly influence the manifestation and severity of apraxia of speech.

The broader implications of this finding are constrained by our focus on chronic measures of both brain integrity and impairment. As a result, we cannot determine whether cerebellar integrity reflects degeneration over time or whether pre-morbid brain health contributes to resilience against cortical injury. While these factors are not mutually exclusive, future studies that incorporate hyperacute imaging alongside both acute and chronic outcomes may help disentangle their contributions. Nonetheless, our current findings highlight that chronic cerebellar integrity is a strong biomarker of long-term impairment.

The stronger association between right cerebellar volume and AOS severity compared to dysarthria further supports the cerebellum’s specific contribution to speech planning rather than execution. Additionally, the cerebellum’s established role in novel motor sequence learning aligns with the characteristic difficulties AOS patients experience with unfamiliar articulation patterns.9,55 These findings extend the current models of AOS, beyond traditional cortical structures, to include the cerebellum as an integral component of the speech motor network. This could have important implications for rehabilitation approaches, suggesting that interventions targeting cerebellar plasticity might be particularly beneficial for individuals with AOS. Future investigations will further refine our understanding of the cerebellum’s role in speech motor control and potentially inform more targeted rehabilitation strategies for individuals with AOS following stroke.

Funding:

This work was supported by the National Institute of Health (P50DC014664, U01DC011739, R01DC008355, RF1MH133701). We would like to acknowledge the participants, students, faculty and staff who have supported the Center for the Study of Aphasia Recovery.

Footnotes

Competing Interests: The authors declare no conflict of interest.

Ethical Considerations: All participants gave informed consent for study participation in accordance with the Declaration of Helsinki.

Data Availability:

The data that support the findings of this study are available from the corresponding author upon reasonable request. Ethics restrict our ability to share identifiable patient data including behavior, health data or scans. Therefore, reasonable request in this instance refers to a request that does not include identifiable data.

References

During the preparation of this work the authors used ChatGPT in order to improve flow and clarity. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

  • 1.Ziegler W et al. The prevalence of apraxia of speech in chronic aphasia after stroke: A bayesian hierarchical analysis. Cortex 151, 15–29 (2022). [DOI] [PubMed] [Google Scholar]
  • 2.Laganaro M Patterns of impairments in AOS and mechanisms of interaction between phonological and phonetic encoding. J Speech Lang Hear Res 55, S1535–43 (2012). [DOI] [PubMed] [Google Scholar]
  • 3.Graff-Radford J et al. The neuroanatomy of pure apraxia of speech in stroke. Brain Lang 129, 43–46 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Levy DF et al. Multivariate lesion symptom mapping for predicting trajectories of recovery from aphasia. Brain Commun 6, fcae024 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Ackermann H, Mathiak K & Riecker A The contribution of the cerebellum to speech production and speech perception: clinical and functional imaging data. Cerebellum 6, 202–213 (2007). [DOI] [PubMed] [Google Scholar]
  • 6.Ackermann H & Hertrich I The contribution of the cerebellum to speech processing. J. Neurolinguistics 13, 95–116 (2000). [Google Scholar]
  • 7.Manni E & Petrosini L A century of cerebellar somatotopy: a debated representation. Nat Rev Neurosci 5, 241–249 (2004). [DOI] [PubMed] [Google Scholar]
  • 8.Salmi J et al. Cognitive and motor loops of the human cerebro-cerebellar system. J Cogn Neurosci 22, 2663–2676 (2010). [DOI] [PubMed] [Google Scholar]
  • 9.Durisko C & Fiez JA Functional activation in the cerebellum during working memory and simple speech tasks. Cortex 46, 896–906 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Desmond JE & Fiez JA Neuroimaging studies of the cerebellum: language, learning and memory. Trends Cogn Sci 2, 355–362 (1998). [DOI] [PubMed] [Google Scholar]
  • 11.Urban PP Speech motor deficits in cerebellar infarctions. Brain Lang 127, 323–326 (2013). [DOI] [PubMed] [Google Scholar]
  • 12.Sebastian R et al. Cerebellar neuromodulation improves naming in post-stroke aphasia. Brain Commun 2, fcaa179 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Urban PP et al. Cerebellar speech representation: lesion topography in dysarthria as derived from cerebellar ischemia and functional magnetic resonance imaging. Arch Neurol 60, 965–972 (2003). [DOI] [PubMed] [Google Scholar]
  • 14.Spencer KA & Slocomb DL The neural basis of ataxic dysarthria. Cerebellum 6, 58–65 (2007). [DOI] [PubMed] [Google Scholar]
  • 15.Hillis AE et al. Re-examining the brain regions crucial for orchestrating speech articulation. Brain 127, 1479–1487 (2004). [DOI] [PubMed] [Google Scholar]
  • 16.Richardson JD, Fillmore P, Rorden C, Lapointe LL & Fridriksson J Re-establishing Broca’s initial findings. Brain Lang 123, 125–130 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Ogar J, Slama H, Dronkers N, Amici S & Gorno-Tempini ML Apraxia of Speech: An overview. Neurocase (2005) doi: 10.1080/13554790500263529. [DOI] [PubMed] [Google Scholar]
  • 18.Kristinsson S et al. Machine learning-based multimodal prediction of language outcomes in chronic aphasia. Hum Brain Mapp 42, 1682–1698 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Kristinsson S et al. Brain Damage Associated with Impaired Sentence Processing in Acute Aphasia. J Cogn Neurosci 32, 256–271 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Matchin W et al. Grammatical Parallelism in Aphasia: A Lesion-Symptom Mapping Study. Neurobiol Lang (Camb) 4, 550–574 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Fridriksson J et al. Anatomy of aphasia revisited. Brain 141, 848–862 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Anglade C, Thiel A & Ansaldo AI The complementary role of the cerebral hemispheres in recovery from aphasia after stroke: a critical review of literature. Brain Inj 28, 138–145 (2014). [DOI] [PubMed] [Google Scholar]
  • 23.Stefaniak JD, Halai AD & Lambon Ralph MA The neural and neurocomputational bases of recovery from post-stroke aphasia. Nat Rev Neurol 16, 43–55 (2020). [DOI] [PubMed] [Google Scholar]
  • 24.Newman-Norlund RD et al. Cerebellar Atrophy and Language Processing in Chronic Left-Hemisphere Stroke. Neurobiol Lang (Camb) 5, 722–735 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Kim JS et al. Degeneration of the Inferior Cerebellar Peduncle After Middle Cerebral Artery Stroke: Another Perspective on Crossed Cerebellar Diaschisis. Stroke 50, 2700–2707 (2019). [DOI] [PubMed] [Google Scholar]
  • 26.Hanakawa T et al. Macrostructural Cerebellar Neuroplasticity Correlates With Motor Recovery After Stroke. Neurorehabil Neural Repair 37, 775–785 (2023). [DOI] [PubMed] [Google Scholar]
  • 27.Wang B, LeBel A & D’Mello AM Ignoring the cerebellum is hindering progress in neuroscience. Trends Cogn Sci 29, 318–330 (2025). [DOI] [PubMed] [Google Scholar]
  • 28.Cerebellar Contributions to Speech and Language. in Neurobiology of Language 73–84 (Academic Press, 2016). [Google Scholar]
  • 29.Duffy JR et al. The Apraxia of Speech Rating Scale: Reliability, Validity, and Utility. Am J Speech Lang Pathol 32, 469–491 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.The apraxia of speech rating scale: A tool for diagnosis and description of apraxia of speech. Journal of Communication Disorders 51, 43–50 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Gibson M et al. The Aphasia Recovery Cohort, an open-source chronic stroke repository. Sci Data 11, 981 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Penny WD, Friston KJ, Ashburner JT, Kiebel SJ & Nichols TE Statistical Parametric Mapping: The Analysis of Functional Brain Images. (Elsevier, 2011). [Google Scholar]
  • 33.Gaser C et al. CAT – A Computational Anatomy Toolbox for the Analysis of Structural MRI Data. Preprint at 10.1101/2022.06.11.495736. [DOI] [PMC free article] [PubMed]
  • 34.Nachev P, Coulthard E, Jäger HR, Kennard C & Husain M Enantiomorphic normalization of focally lesioned brains. Neuroimage 39, 1215–1226 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Yourganov G, Fridriksson J, Stark B & Rorden C Removal of artifacts from restingstate fMRI data in stroke. Neuroimage Clin 17, 297–305 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Whitwell JL, Crum WR, Watt HC & Fox NC Normalization of cerebral volumes by use of intracranial volume: implications for longitudinal quantitative MR imaging. AJNR Am. J. Neuroradiol 22, 1483–1489 (2001). [PMC free article] [PubMed] [Google Scholar]
  • 37.Diedrichsen J, Balsters JH, Flavell J, Cussans E & Ramnani N A probabilistic MR atlas of the human cerebellum. Neuroimage 46, 39–46 (2009). [DOI] [PubMed] [Google Scholar]
  • 38.Rorden C & Brett M Stereotaxic Display of Brain Lesions. Behavioural Neurology 12, 191–200 (2000). [DOI] [PubMed] [Google Scholar]
  • 39.West RM Best practice in statistics: The use of log transformation. Ann Clin Biochem 59, 162–165 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Lin DDM et al. Crossed cerebellar diaschisis in acute stroke detected by dynamic susceptibility contrast MR perfusion imaging. AJNR Am J Neuroradiol 30, 710–715 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Liu C-F et al. Digital 3D Brain MRI Arterial Territories Atlas. Sci Data 10, 74 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Fridriksson J, Fillmore P, Guo D & Rorden C Chronic Broca’s Aphasia Is Caused by Damage to Broca’s and Wernicke’s Areas. Cereb Cortex 25, 4689–4696 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Fridriksson J, Baker JM & Moser D Cortical mapping of naming errors in aphasia. Hum Brain Mapp 30, 2487–2498 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Matchin W et al. Functional differentiation in the language network revealed by lesionsymptom mapping. Neuroimage 247, 118778 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Fridriksson J, Richardson JD, Fillmore P & Cai B Left hemisphere plasticity and aphasia recovery. Neuroimage 60, 854–863 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Thompson CK, den Ouden D-B, Bonakdarpour B, Garibaldi K & Parrish TB Neural plasticity and treatment-induced recovery of sentence processing in agrammatism. Neuropsychologia 48, 3211–3227 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Heiss WD, Kessler J, Thiel A, Ghaemi M & Karbe H Differential capacity of left and right hemispheric areas for compensation of poststroke aphasia. Ann Neurol 45, 430–438 (1999). [DOI] [PubMed] [Google Scholar]
  • 48.Behroozmand R, Bonilha L, Rorden C, Hickok G & Fridriksson J Neural correlates of impaired vocal feedback control in post-stroke aphasia. Neuroimage 250, 118938 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Arleo A et al. Consensus Paper: Cerebellum and Ageing. Cerebellum 23, 802–832 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Glover GH Overview of functional magnetic resonance imaging. Neurosurg Clin N Am 22, 133–9, vii (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Ganzetti M, Wenderoth N & Mantini D Mapping pathological changes in brain structure by combining T1- and T2-weighted MR imaging data. Neuroradiology 57, 917–928 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Kamali A, Kramer LA, Frye RE, Butler IJ & Hasan KM Diffusion tensor tractography of the human brain cortico-ponto-cerebellar pathways: a quantitative preliminary study. J Magn Reson Imaging 32, 809–817 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Palesi F et al. Contralateral cortico-ponto-cerebellar pathways reconstruction in humans in vivo: implications for reciprocal cerebro-cerebellar structural connectivity in motor and non-motor areas. Sci Rep 7, 12841 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Diedrichsen J A spatially unbiased atlas template of the human cerebellum. Neuroimage 33, 127–138 (2006). [DOI] [PubMed] [Google Scholar]
  • 55.Lundin NB et al. Neural evidence of switch processes during semantic and phonetic foraging in human memory. Proc Natl Acad Sci U S A 120, e2312462120 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request. Ethics restrict our ability to share identifiable patient data including behavior, health data or scans. Therefore, reasonable request in this instance refers to a request that does not include identifiable data.

RESOURCES